In our previous study 19, we characterized several alternatively spliced transcripts harbouring exons 1 to 5 of the original PMCHL1/PMCHL2 genes. Six transcripts corresponding to PMCHL1 spliced RNA were found in testis and/or fetal brain. Two PMCHL2 spliced RNA were reported only in testis. In order to further precise the exon/intron structure of PMCHL genes and to further investigate the tissue distribution of spliced PMCHL transcripts, we examined the presence of additional RNAs in human testis and cortex Marathon cDNA libraries, as well as in a human fetal prefrontal cortex sample. For this, we performed PCR experiments (Figure 1) using primer pairs designed to amplify (in one or two rounds of PCR) transcripts encompassing the most distant known exons, previously named exons 1 and 5. Sub-cloning of the PCR products and sequencing of individual clones allowed the discovery of a novel exon, located between exon 2 and former exon 3. It was named exon 3 and previously named exons 3, 4 and 5 are now renamed exons 4, 5 and 6 (Figure 1A, top panel).

We identified six PMCHL1 spliced variants in adult testicular Marathon cDNA library (Figure 1B), two of which [GenBank:EU921424, GenBank:EU921428] had already been identified in the same cDNA library [GenBank:AY008408, GenBank:AY008410, respectively]. However, transcript [GenBank:EU921424] bears one A to G mutation within exon 1, resulting in an arginine to glycine mutation in ORF1. Noteworthy, transcript [GenBank:EU921428] corresponds also to two testis ESTs (IMAGE clone 1753807 [GenBank:AI203691; EMBL:BX091674] and IMAGE clone 1326573 [GenBank:AA724728]), and was also found in a human fetal brain Marathon cDNA library in our previous study 19. This apparently abundant transcript harbours exons 1-2b and 6a, like most spliced transcripts in testis which were obtained in a single round of PCR. In contrast, transcripts harbouring exon 6b were identified after two rounds of PCR and never contained exon 6a. Transcripts containing exon 2a were never observed in the present study.

We also identified four novel alternative PMCHL2 splicings in adult testis, which all contained a partial exon 3 (3b₂) shorter than the original exon 3 (3b₁) observed in a PMCHL1 transcript (Figure 1B). Sequence analysis (see Figure 2) revealed that mutations in the PMCHL2 sequence created a novel gene-specific splice donor site (3b₂) which is systematically used in the present PMCHL2 transcripts.

We further identified one novel PMCHL1 splice variant in a fetal prefrontal cortex sample (Figure 1B). This transcript was the only one to harbour an alternative splice donor in exon 4, which was never observed in testis RNAs.

In our previous study 19, we reported two alternative acceptor sites in exon 5b (previously named exon 4b) separated by only four nucleotides and with apparent testis- and brain-specificities (indicated by superscript t/b in Figure 1). Our present results show that most testis transcripts use the exon 5a splice acceptor site, and two use the alternative 5b_t site. However, the PMCHL1 transcripts identified in fetal brain also use the alternative 5b_t site indicating that it could not be considered anymore as a testis-specific splice acceptor site. The alternative 5b_b site, previously reported in a fetal brain transcript was not found in our present study.

In contrast to human testis and fetal brain, we could not detect any spliced PMCHL RNA harbouring exons 1 and 6a in the human adult cortex Marathon cDNA library using the 3–7/3–30 primer pair. This primer pair was further used in RT-PCR experiments combined with Southern blot to determine the tissue distribution of the spliced transcripts in testis, prefrontal cortex and cerebellum in adult human and macaque (Figure 1C). In agreement with our results using the Marathon cDNA libraries, we detected spliced PMCHL RNA harbouring exons 1 and 6a in human adult testis, but not in adult prefrontal cortex and cerebellum. Thus, PMCHL1 transcripts are found in testis and fetal brain and are more abundant than PMCHL2 transcripts that are observed only in testis. In addition, PMCHL2 gene expression was reported in HT1080 cells subjected to RAGE (random activation of gene expression), in which an EST [GenBank:BG184695] encompassing exons 2a and 3 of PMCHL2 has been identified.

In macaque, no spliced transcripts were identified by Southern blot (Figure 1C) in agreement with our sequence analysis indicating that the macaque PMCHL1 gene lacks the exon 6a acceptor splice site (see below).

Taken together, our findings indicate that PMCHL1 and PMCHL2 genes: i) give rise to a complex pattern of alternative splicings, ii) are subject to distinct tissue-specific expressions and iii) are developmentally regulated (i.e. expressed in fetal but not adult cortex).

The finding that a rather high diversity of spliced transcripts are present in testis is not surprising, because of the permissive chromatin environment present in gonads, allowing high transcriptional activity even from weak tissue-specific promoters 29. Thus, most retroposons evolve into non-functional pseudogenes that are transcribed only in the testis 30. However, the abrupt emergence of a new chimeric gene in primates could potentially contribute to reproductive barriers and thus play a role in speciation 31. In this regard the hominoid-specific oncogene Tre2 appears expressed only in testis while the two parental genes USP32 and TBC1D3, that fused to generate the Tre2 gene, are expressed in a broad range of human tissues 32. In addition, the presence of spliced PMCHL transcripts in fetal brain, is rather suggestive of a functional role during human brain development. This would imply that the retroposon acquired an active promoter and has been subjected to selection pressure. Whether these spliced PMCHL transcripts actually play a functional role in testis and fetal brain is an obvious question, which we further addressed below.

We next addressed the protein coding potential of the spliced PMCHL RNAs. We examined the sequences of all PMCHL transcripts reported in the present and in our previous study 19 to identify ORFs longer than 100 bp. Ten short ORFs of less than 300 bp were found. PMCHL transcripts harbouring exons 5-6a or 5-6b present all together seven ORFs that are 120 to 198 bp in length (Figure 5), i.e. they would encode proteins of 40 to 66 amino acids. ORFs of less than 300 bp (i.e. 100 amino acids) are often assumed not to be translated. However, many well known functional proteins of less than 100 amino acids in length have been reported, including the small inducible cytokine families CCL and CXCL 48, and the xenobiotic defensin and defensin-related cryptidin factors 49. Furthermore, a recent study has shown that among the 31,035 predicted proteins encoded by the 102,801 FANTOM mouse full-length cDNA sequences, 12% of the proteins (i.e. 1,683 proteins) are less than 100 amino acids in length 50. This suggests that there might be up to 4 times more small proteins than the 424 entries present for Mus musculus to date in the SwissProt protein database (release 56). Interestingly, most of the small proteins with known function are evolutionarily conserved 48 or present conserved sequence motifs 49. Notably, a recent report 51 indicates that ORFs < 300 bp in length, that are not evolutionarily conserved, are unlikely to be translated into functional proteins. Given that the PMCHL ORFs present on exons 5-6a and 5-6b are not conserved among Homo sapiens, Pan troglodytes and Pongo pygmaeus, due to frameshift-causing insertions/deletions (Figure 5), we propose that these ORFs are most likely non-functional.

The longest ORF identified on PMCHL transcripts is 294 bp (98 amino acids) long and locates within the Alu sequence in exon 2a. Four transcripts harbouring exon 2a were identified in testis in our previous study 19. However, the corresponding putative protein is not conserved due to a single nucleotide insertion in the human PMCHL1 sequence causing a frameshift in the middle of the ORF. Therefore, this ORF is also spurious according to the criteria of Clamp and colleagues 51.

PMCHL transcripts encompassing exons 1-2a and 1-2b harbour two ORFs, named ORF1a and ORF1b, respectively (Figure 6A, B, C). Even though these ORFs are also less than 300 bp in length, they present the same lengths in Homo sapiens, Pan troglodytes and Pongo pygmaeus, and share > 90% sequence identity. In Macaca fascicularis, the ORF is shortened and runs only on exon 1 due to the presence of an early stop cordon. These ORFs are of particular interest because they share a large part of sequence identity with ORF1 present on unspliced PMCHL RNA, and with the pro-MCH precursor because it mainly locates in exon 1, i.e. in the region derived from exon 2 of the ancestor PMCH gene. The putative 8 kD protein corresponding to ORF1, previously named VMCH-p8, presents a putative nuclear localisation signal (NLS) at the N-terminus (KPKKK, shaded in grey in Figure 6B), and is among the longest ORFs, encoding 72 amino acids (Figure 6A, B, C). In a previous study 27, we examined the protein coding potential of ORF1 carrying out in vitro translation experiments and COS-7 cell transfections with the Flag epitope-tagged ORF1. The results indicated a weak protein-coding potential, depending on particular plasmid constructions, providing mRNA stabilising elements and enhanced promoter activity 27. In the present study, we used a VMCH-p8 antiserum directed against the thirteen N-terminal VMCH-p8 amino acids, comprising the putative NLS (see Figure 6B). This allows the determination of the expression of ORF1, as well as the ORF1a and ORF1b variants (sharing the N-terminal epitope). The reactivity of the VMCH-p8 antiserum was demonstrated in Western blot experiments using a recombinant GST-VMCH-p8 protein produced in bacteria. VMCH-p8 antiserum recognized the GST-VMCH-p8 protein, migrating at about 34 kD, whereas the preimmune serum did not (Figure 6D). Next, we used the VMCH-p8 antiserum to examine expression of ORF1 and its variants in human and macaque tissues (Figure 6E). We tested human adult testis, hippocampus and prefrontal cortex extracts from a new-born and a foetus, as well as four Macaca fascicularis cerebral areas (supplementary motor area, cerebellum, prefrontal cortex and visual area). These tissues and cerebral areas were chosen for the presence of ORF1-bearing PMCHL transcripts in RT-PCR experiments (2728; our unpublished data). In our Western blot experiments, no signal could be detected in all human and macaque tissues tested, at the expected size of 8–9 kD for the putative VMCH-p8 protein and its variants. This strongly suggests that these putative proteins are not translated in vivo in the human and macaque tissues that we tested. We further carried out Western blot and immunoprecipitation experiments on HEK293 cells transfected with PMCHL1/2 sequences bearing ORF1 to detect low levels of VMCH-p8 protein. Even though high levels of ORF1-bearing PMCHL1/2 transcripts were detected by RT-PCR, no signal corresponding to the VMCH-p8 protein could be detected (data not shown). One explanation for the lack of protein detection, that we cannot exclude, is a very low protein expression level below our detection threshold. Also, for the putative Macaque protein, we further cannot exclude an altered epitope-recognition of the antibody due to a lysine to glutamic acid mutation within the epitope. Assuming that the failure to detect the VMCH-p8 protein or its variants is not due to these technical limitations, the lack of translation of ORF1 like-bearing mRNAs could reside in the moderate consensus with the optimal sequence for translation initiation described by Kozak 52. Actually, only the consensual adenine at position -3 is present.

What might be the role(s) of the large variety of spliced PMCHL mRNAs in human testis and fetal brain? It is tempting to propose that these PMCHL transcripts work mainly as an mRNA-like non-protein coding RNA (ncRNA). Since the realization that 98% of the transcriptional output in mammals consists of ncRNAs, the enthusiasm for this class of RNAs has grown tremendously 535455 and has been granted its own NONCODE database 56. Numerous classes of ncRNAs have been reported, most of which are small ncRNAs (including miRNAs, siRNAs and snoRNAs), but also long ncRNAs (ranging from 1 to more than 100kb) such as Xist and the antisense Tsix transcripts involved in × inactivation in mammals 5758, or the Air RNA that appears to be responsible for imprinted repression of nearby genes (including Igf2r gene) through an antisense-mediated mechanism 59. Several mRNA-like ncRNAs that are transcribed by RNA polymerase II, spliced and polyadenylated have also been reported 606162, including in human 6364. Interestingly, many small ncRNAs are located in introns of coding or non-coding mRNAs 5465. The functional roles of ncRNA are diverse, corresponding mainly to adaptor functions targeting nucleic acids to various enzymatic complexes (involved in RNA processing, splicing, transcription...) and gene expression regulation/silencing (involved in virtually all cellular functions).

Do the PMCHL transcripts host small ncRNAs in their introns, and/or do the PMCHL transcripts control the expression of neighbouring genes in cis (an obvious candidate is the CDH12 gene) or in trans through RNA-RNA duplexes (obvious candidates are the PMCH and Antisense RNA Overlapping MCH (AROM) genes)? We are now addressing these intriguing questions.

Human prefrontal cortex and cerebellum from adults were provided by the National Neurological Research Specimen Bank (Los Angeles, CA, USA) and by the GIE Neuro-CEB (Hôpital de la Pitié-Salpétrière, Paris, France), which collect tissues with the full authorization of the respective local ethical committees. Human adult testis RNA was purchased from BioChain/Cliniscience (France). Dr. A. Coquerel (CHU Rouen, France) provided human prefrontal cortex and hippocampus tissues from a newborn. Dr. D. Jordan (Faculté de médecine, Lyon, France) provided human prefrontal cortex and hippocampus tissues from a foetus. Collection of human new-born and foetal tissues was according to the french legislation of parental consent and with the approval of local ethical committees. Human adult testis total proteins were purchased from BioChain/Cliniscience (France). Testis, prefrontal cortex, cerebellum, visual area, and supplementary motor area samples from three adult macaques (Macaca fascicularis) were obtained from Dr. E. Bezard at the Biothèque Primate/Primatech (CNRS, Bordeaux, France), where tissue collection is carried out in agreement with the European Communities Council Directive of November 24, 1986 (86/609/EEC).

Genomic DNAs were collected from Cebus capucinus (gift from B. Dutrillaux, cytogénétique moléculaire et oncologie, CNRS, Institut Curie, Paris, France), Tarsius syrichta, Saguinus oedipus, Cebus apella, Chlorocebus aethiops, Hylobates lar, Pan paniscus and Pan troglodytes (gift from Dr P. Dijan, CEREMOD, Meudon, France) and were already used in previous studies 28. Other genomic samples were kindly provided by San Diego Zoo/CRES (Pan paniscus, Gorilla g.g., Pongo pygmaeus, Pongo p.abelii, Hylobates lar, Macaca silenus) and by Prof A. Blancher (Rangueil Hospital, Toulouse, France) (Cebus appella, Pan troglodytes). Genomic DNA was isolated from the occipital cortex of a Macaca fascicularis (provided by Dr. E. Bezard (Biothèque Primate/Primatech, CNRS, Bordeaux, France) according to the Blin and Stafford's method 66.

Total RNAs were extracted from human and macaque tissues according to standard guanidium phenol method 67 and using a FastPrep apparatus (FP220A Thermo instrument, Qbiogene, France). Contaminating genomic DNA was removed from RNA preparation by RQ1 RNase-free DNase treatment (Promega) according to the manufacturer's protocol. cDNAs were synthesized by reverse-transcription (RT) of 2 μg of total DNase-treated RNAs using the SuperScript TM II Reverse Transcriptase (Invitrogen) and oligo dT according to the manufacturer's protocol.

Oligonucleotides (list provided in Table 1) were purchased from Eurogentec (Belgium).

For genomic DNA, 100–200 ng were PCR-amplified using the oligonucleotide couples indicated in Table 1 and the LA Taq polymerase (Takara) following the supplier's protocol. Thirty-five cycles of amplification were carried out as follows: 30 s at 94°C (denaturation), 30 s at annealing temperature (indicated in Table 1), 1 to 10 min at 72°C (extension). A final extension step of 7 min at 72°C was performed. PCR products were purified using the NucleoSpin kit (Machery Nagel) and sequenced.

For RT samples and Marathon cDNA libraries, 2 μl were PCR-amplified using the indicated primer pairs and the HotMaster Taq DNA polymerase (Eppendorf) following the supplier's protocol. Thirty-eight cycles of amplification were carried out as follows: 30 s at 94°C (denaturation), 30 s at annealing temperature (indicated in Table 1), 2 min at 65°C (extension). A final extension step of 7 min at 72°C was performed. When necessary, nested PCR was performed with internal primers using 2 μl of a 1:20 dilution of the first round products. PCR-amplified fragments were subcloned into the pGEM-T Easy vector (Promega) and transfected into TOP10 thermocompetent cells (Invitrogen) according to the manufacturer's instructions, followed by plasmid DNA preparation using a Qiaprep Spin Miniprep kit (Qiagen) and sequencing.

PCR products obtained with primer pair 3–7/3–30 (thirty-five cycles) were electrophoresed on a 1% agarose gel containing ethidium bromide and were visualized under UV. The gel was then denatured 15 min in 500 mM NaOH, 1.5 M NaCl solution, neutralized 15 min in 500 mM Tris, 1.5 M NaCl, and soaked 5 min in 2 × SCC solution (300 mM NaCl, 30 mM sodium citrate). The DNA was transferred overnight as a gravity-dry blot onto a cellulose membrane (Biodyne B, Pall Corporation, FL, USA). The membrane was prehybridized for 4 h at 65°C in Church solution (500 mM Na₂HPO₄, pH 6.8, 5% SDS), hybridized overnight at 65°C in fresh Church solution containing previously prepared ³²P-labeled PMCHL-specific probe 1 corresponding to the fragment amplified with primer pair 3–25/3–30 (see Table 1) at 5.10⁵ dpm.ml^-1. ³²P-labeled probes were prepared using the Prime-a-gene labelling system (Promega) according to the manufacturer's protocol. After hybridization, the membrane was washed twice 15 min in 2 × SSPE and twice 10 min in 1 × SSPE. Hybridized radioactive probes were detected with a Fujifilm phosphoimager (FLA-5100).

Sequencing of PCR-amplified fragments was carried out on both DNA strands using the Ampli Taq Polymerase FS, the Big Dye Terminator 1.1 sequencing kit (Applera), and a ABI PRISM 3100 sequencer (Perkin Elmer). Sequences obtained from the public databases (EMBL/GenBank/DDBJ) and fragments sequenced by PCR were aligned manually using SEAVIEW 68. Species are: Homo sapiens (HSA), Pan paniscus (PPA), Pan troglodytes (PTR), Pongo pygmaeus (PPY), Hylobates lar (HLA), Macaca mulatta (MML), Macaca fascicularis (MFA), Cercopithecus hamlyni (CHA), Chlorocebus aethiops (CAE), Cebus capucinus (CCA), Cebus apella (CAP), Saginus oedipus (SOE), Tarsius syrichta (TSY).

Phylogenetic dendrograms were reconstructed according to three different methods: Neighbour Joining (BIONJ), Maximum Likelihood (ML, using the Global option), and Maximum Parsimony (MP). For the Neighbour Joining (NJ) analysis, a distance matrix was calculated by DNADIST according to the Kimura two parameters correction. Bootstraps were done using 1,000 replications, BIONJ and Kimura two parameters correction. BIONJ was according to Gascuel 69, ML and MP were from PHYLIP (Phylogeny Inference Package, version 3.573c, distributed by J. Felsenstein, Department of Genetics, UW, Seattle, WA, USA). Phylogenetic analyses were done excluding domains that were not common to every sequence as well as low complexity domains that could not be properly aligned. The phylogenetic dendrograms were drawn using NJPLOT 70.

Human and macaque tissues were homogenized in RIPA buffer (20 mM Tris HCl, pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholic acid, 0.1% SDS, 2 mM EDTA, protease inhibitor cocktail Complete (Roche)) using a FastPrep apparatus (FP220A Thermo instrument, Qbiogene, France), incubated on ice for 30 min, and centrifuged at 20000 × g for 15 min at 4°C. Proteins in the supernatants were quantitated using a commercial Bradford reagent (BioRad).

PMCHL1 ORF1 encoding the putative VMCH-p8 protein was sub-cloned into the BamH1/EcoR1 sites of the pGEX-3X vector, in frame with GST (Amersham Biosciences). The construct was used to transform thermocompetent Rosetta cells (Novagen) and the recombinant GST-VMCH-p8 protein was produced and purified using glutathione-sepharose (Amersham Biosciences) beads according to the manufacturer's instructions.

Polyclonal antibodies were raised against the putative VMCH-p8 protein encoded by PMCHL1 ORF1. A peptide comprising the thirteen N-terminal amino acids of the sequence (MLSQKPKKKHNFL) was designed by Dr B. Cardinaud (IPMC, Valbonne, France), synthesized and coupled to keyhole limpet haemocyanin (KLH) before rabbit immunization (Genaxis, Nîmes, France). Anti-VMCH-p8 antiserum was used at a final dilution of 1:1,000. Secondary HRP-coupled goat anti-rabbit antibodies (Jackson ImmunoResearch) were used at a 1:10,000 dilution.

Proteins were separated on 12% Tris-glycine or 16.5% Tris-tricine gels under reducing conditions and transferred to nitrocellulose membranes (Schleicher & Schuell, Germany) using a wet tank transfer system (BioRad). Membranes were blocked 1 h in TBS-T (137 mM NaCl, 2.7 mM KCl, 2.5 mM Tris, pH 7.4, 0.1% Tween-20) containing 5% fetal calf serum, incubated for 2 h at room temperature with primary antibodies (1:1000 dilution), followed by 1 h incubation with secondary antibodies, and revealed with the SuperSignal West Pico (Pierce) chemiluminescence detection system.